Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
130817
CA 02470687 2004-06-10
METHOD, SYSTEM AND COMPUTER PRODUCT FOR ESTIMATING A
REMAINING EQUIPMENT LIFE
BACKGROUND OF THE INVENTION
The invention generally relates to estimating a remaining equipment life and
more
specifically to a method and system for estimating a remaining equipment life
based
on multiple parameters.
Equipment life estimates are usually performed for estimating a remaining
equipment
life and are also useful in the determination of the time to failure and
reliability of
equipment components. Typically, age based population distributions serve as a
primary information source for estimating a remaining equipment life. In that
case,
the current age of the equipment is taken as an indication of the time to
failure of the
equipment. However, age based population distributions have considerable
degrees
of variability in their distribution which reduces the usefulness of the
remaining
equipment life estimates.
BRIEF DESCRIPTION OF THE INVENTION
One technique to address the need of determining a more refined estimate of
the
remaining equipment life is to integrate other (possibly heterogeneous)
information
sources that are potential indicators of the remaining equipment life.
In one embodiment, a method and computer readable medium for estimating a
remaining equipment life based on a plurality of parameters is provided. Data
relating
to the plurality of parameters are collected. The data are stored and
integrated. The
integrated data are used to estimate the remaining equipment life.
In a second embodiment, a system for estimating a remaining equipment life
based on
a plurality of parameters is provided. The system comprises a data storage
component
configured to store data relating to the plurality of parameters, a data
integration
1
130817
CA 02470687 2004-06-10
component configured to integrate the stored data and a life estimation
component
configured to estimate the remaining equipment life using the integrated data.
BRIEF DESCRIPTION OF THE DRAWINGS
Fig. 1 illustrates a top-level component architecture diagram of an equipment
life
estimation system for estimating a remaining equipment life;
Fig. 2 is a block diagram illustrating the various data sources used by the
equipment
life estimation system of Fig. 1 in the estimation of the remaining equipment
life;
Fig. 3 is an illustrated embodiment of the steps performed by the equipment
life
estimation system of Fig. 1 to estimate the remaining equipment life;
Figs. 4-5 are graphs illustrating Weibull curves derived for two equipment
components using a base equipment age;
Fig. 6 is a table illustrating the goodness of fit of the graphs in Figs 4-5;
Figs 7-8 are graphs illustrating Weibull curves derived for two equipment
components
based on a unified age adjustment value; and
Fig. 9 is a table illustrating the comparison of the goodness of fit of the
graphs of
Figs. 4-S vs. the graphs in Figs.7-8.
DETAILED DESCRIPTION OF THE INVENTION
Fig. 1 illustrates a top-level component architecture diagram of an equipment
life
estimation system 30 for estimating a remaining equipment life. In accordance
with
one embodiment of the invention, system 30 comprises data sources 40, a data
storage
component 50, a data integration component 60 comprising a data modeling
subcomponent 70 and a data mapping subcomponent 80 and a life estimation
component 90. Each component is described in further detail below.
Data sources 40 are used by the life estimation system 30 in the estimation of
the
remaining equipment life. An efficient method far estimating remaining
equipment
2
130817
CA 02470687 2004-06-10
life would be to consider these various heterogeneous data sources of
information in
the estimation of the remaining equipment life. The data sources 40 comprise
sources
such as, for example, failure events, fault code data, usage pattern data, age
data, test
results, maintenance practices, Weibull curves and heuristics related to
equipment
components. In a specific embodiment of the invention, the data sources 40
comprise
usage pattern data, fault code data and age data related to equipment
components.
Fig. 2 describes the data sources 40 used by the equipment life estimation
system of
Fig. 1 in further detail.
Continuing with reference to Fig. 1, the data storage component 50 is
configured to
store data received from data sources 40. In one embodiment, the data storage
component is represented by a relational database of tables.
The data integration component 60 is configured to integrate the data stored
in the
data storage component 50. The data integration component 60 further comprises
a
data modeling subcomponent 70. 'The data modeling subcomponent 70 is
configured
to model a plurality of relationships relevant to a plurality of parameters
represented
by the data sources 40. The modeling comprises enumerating the data sources 40
stored in the data storage component 50, and modeling relationships between
the
parameters represented by the data sources. The data, modeling subcomponent 70
categorizes the data sources 40 based on their cornrnon properties. Common
properties comprise properties that are common to all equipment components and
facilitate the comparison of the parameters represented by the heterogeneous
data
sources from a unified standpoint. In a specific embodiment of the invention,
the
common property is derived based on transforming or mapping the parameters
represented by the data sources into an age adjustment state (indicative of an
impact
on a wear state) related to the equipment components. 'The categorization of
the data
sources 40 is accomplished using an ontological representation or other common
representation mechanism of the data sources. The data integration component
60
further comprises a data mapping subcomponent 80. The data mapping
subcomponent 80 is configured to determine a representation for the parameters
in
terms of a unified index indicative of the remaining equipment life. In a
specific
3
130817
CA 02470687 2004-06-10
embodiment of the invention, the unified index is referred to as an age
adjustment
index. The age adjustment index corresponds to an age adjustment value for
refinement of the remaining equipment life estimate. The basis for age
adjustment is
that the age of an equipment component can be adjusted or refined based on the
wear
that the equipment component is exposed to. That is, the knowledge about the
wear
state of an equipment component provides a more refined life estimate. In a
more
specific embodiment of the invention, the wear state is defined or specified
in terms
of the usage pattern data and fault code data parameters related to equipment
components.
The data mapping subcomponent 80 transforms or maps the parameters to the age
adjustment index. The data mapping subcomponent 80 estimates the age
adjustment
state from the above transformations and fuses the age data related to the
equipment
component with the age adjustment state to arrive at a 'unified age adjustment
value.
The life estimation component 90 then estimates the remaining equipment life
based
on the unified age adjustment value. Fig. 3 describes in. further detail the
set of steps
performed by the equipment life estimation system 30 to estimate the remaining
equipment life. In a specific embodiment, the equipment life estimation system
30 is
used to estimate the remaining life of vehicle or locomotive components. In a
more
specific embodiment, the locomotive components comprise the power assembly and
turbo charger.
Fig. 2 is a block diagram illustrating the various data sources that may be
used by the
equipment life estimation system of Fig. 1 in the estimation of the remaining
equipment life. The various data sources represent potential indicators or
parameters
in the estimation of the remaining equipment life and serve to determine a
more
refined remaining life estimate. A more refined life estimate that would
enable
limiting the use of equipment components that have high wear and high risk of
failure
and maximize the use of equipment components with low wear and lower failure
risks
and in turn improve the cost associated with maintenance operations for these
equipment components. These data sources are compared from a unified
standpoint
and integrated.
4
130817
CA 02470687 2004-06-10
The data sources comprise failure events 204, fault code: data 202, usage
pattern data
208, age data 210, test results 206, maintenance practices 212, Weibull curves
200
and heuristics 214. Failure events 204 provide information about the failure
of
specific equipment components. In one embodiment, an equipment component that
has failed numerous times indicates exposure to severe v~~ear. In another
embodiment,
fault code data 202 is indicative of heavy wear. Fault codes report
information about
component overload, component overheating, etc., indicative of the wear of the
equipment component. Usage pattern data 208 provide information about specific
load conditions subjected to by equipment components. Usage pattern data 208
also
indicate the time duration for which an equipment component was subjected to a
particular load and for how Lang. Information on age data 210 is an indicator
for
average wear of equipment components for a given age.. Typically, component
wear
increases sharply at the beginning of its life, then settles to a more
moderate wear
slope, and increases again towards the end of its life. Weibull curves 200
provide
information regarding equipment life distributions. Test results 206 report
about
specific wear related parameters such as dimensional changes or operational
behavior
acquired during inspections. Maintenance practices 212 indicate differences in
maintenance practices among various equipment components that affect the wear.
Heuristics 214 indicate experience-derived knowledge about a set of data
sources that
describes component behavior relating to wear in more complex relations using
linguistic rules. The above various data sources are then. fused together in
step 216, in
order to estimate the remaining equipment life in step 218. Step 216 is
described in
further detail in step 310 and step 218 is described in further detail in step
312.
~ne of ordinary skill in the art will recogxiize that the above listing of
data sources is
for illustrative purposes and is not meant to limit other types of data
sources that can
be used by the equipment life estimation system 30 in the estimation of the
remaining
equipment life.
Fig. 3 is an illustrated embodiment of the steps performed by the equipment
life
estimation system of Fig. 1 to estimate the remaining equipment life. As
shown, the
130817
CA 02470687 2004-06-10
process starts in step 300 and then passes to step 302. Each step is described
in
further detail below.
In step 302, data related to the age, fault codes, and usage pattern
parameters are
collected from the data sources 40 and stored in the data storage component
50. The
data storage component, represented by the relational database comprises
fields and
methods. The field and methods comprise information pertaining to specific
equipment components. In one embodiment, the fields specify parameters, such
as
usage pattern data and fault code data related to the equipment component,
that are to
be stored in the database and the methods specify commands used to retrieve
the data
related to the parameters.
In step 304, the fault code data is mapped to a fault code to age adjustment
index. In
one embodiment of the invention, the mapping of fault code to age adjustment
index
comprises representing the fault code as the number of error messages or error
log
entries generated by the equipment. Error messages could be generated due to
overheating of the equipment component, for example. The number of error
messages is considered to be related to the impact on th.e wear of the
equipment. The
larger the number of error messages generated by the equipment, the larger is
the
impact on the weax, and hence the age adjustment index of the equipment.
In step 306, the usage pattern data is mapped to a usage to age adjustment
index. In
one embodiment of the invention, the mapping of usage to age adjustment index
is
based on the number of megawatt hours consumed by the equipment component and
comprises representing the usage pattern data as a ratio of a weighted average
of the
time spent at a plurality of Load settings relevant to the equipment component
to the
power value consumed by the equipment component. The plurality of load
settings
are indicative of a type and duration of a plurality of load conditions
subjected to by
the equipment component. An equipment component whose usage is high will have
a
larger impact on the wear, and hence the age adjustment index of the
equipment.
In step 308, the age adjustment state is estimated from the fault code to age
adjustment index and the usage to age adjustment index derived in steps 304
and 306
6
130817
CA 02470687 2004-06-10
respectively. The estimating comprises calculating the fault code to age
adjustment
index and the usage to age adjustment index. The fault code to age adjustment
index
is calculated using a suitable nonlinear squashing function. One embodiment of
this
function is:
age _ adjustment _ indexe,~,_r°gs ° 2 ~ 1 + a ~"°'#
i'r°r r°gr-~ ~ ~.5~ (
Here, a,° scales the slope of the curve and is a tunable parameter and
(3e is a positional
parameter, and is also tunable. Similarly, the usage pattern to age adjustment
index is
calculated using a suitable nonlinear squashing function. One embodiment of
this
function is:
age _ adjustment _ index"5°&e- pattern 2C 1 + a (aa ~u'~lge-'ndex-(t"
(2)
Here, a" scales the slope of the curve and is a tunable parameter and Vii" is
a positional
parameter, and is also tunable.
The parameters a and (3 are used to tune the mapping equations (1) and (2).
The
tuning comprises using a suitable optimization function. One embodiment of an
optimization function is a genetic algorithm that determines the set of
parameters
based on a fitness function such as the sum-squared error and the percentage
of data
points within a pre-defined confidence interval. In one embodiment of the
invention,
a 95% confidence interval of the Weibull curve is used in the fitness
function.
Equations (1) and (2) are derived based on standard non-linear squashing
functions.
Non-linear squashing functions are made use of in machine learning methods
such as
neural networks. Non-lineax squashing functions are monotonically increasing
functions that take as input, values between -oo and +~ and return values in a
finite
interval. The mapping equations (1) and (2) take quantified inputs such as
usage in
megawatt hours consumed or number of error Iog entries and return a continuous
number between -1 and +1 representing the age adjustment index, as output.
Continuing with the flow chart of Figure 3, in step 310, the age data of the
equipment
is fused with the age adjustment state derived from the equations (1) and (2)
of step
7
130817
CA 02470687 2004-06-10
308. The fusion results in the determination of a unified age adjustment
value.
Various strategies exist to accomplish the fusion in step 310. The fusion uses
an
aggregation technique to estimate the remaining equipment life. In one
embodiment
of the invention, the fusion technique used is a linear aggregation technique.
In
another embodiment, the fusion technique used is a non-linear aggregation
technique.
Different weights are assigned to the calculated age adjustment indices of
step 308.
Then the age adjustment state, comprising the weighted age adjustment indices,
is
fused with the age data to determine a unified age adjustment value. The
fusion
computes a weighted sum of the age data and the age adjustment state. In step
312, a
unified age adjustment value indicative of the remaining equipment life is
determined
using the following equation:
unified _ age _ adjustment _ value = age + C2 * age _ adjustment _ index~age_
porter" + C3 * age _ adjustment _ index ~r_,o~,
(3)
Here, C2*age_ _adjustment_lndeXusage_pattern and C3'kage_adjustment
index,.,.°r logs
represent the weighted age adjustment indices, respectively. The weights C2
and C3
indicate a degree of emphasis placed on each of the weighted age adjustment
indices.
C2 and C3 are tuned using a suitable optimization function. One embodiment of
an
optimization function is a genetic algorithm that determines the set of
parameters
based on a fitness function such as the sum-squared error and the percentage
of data
points within a pre-defined confidence interval. In one embodiment of the
invention,
a 95% confidence interval of the Weibull curve is used in the fitness
function.
The following figures illustrate the Weibull curves derived for equipment
components
in one embodiment of the invention. Weibull distrihutions are generally used
to
model various life distributions and in the determination of reliability of
equipment
components. Reliability is defined as the probability of failure of an
equipment
component at a specified period of time. In a specific embodiment of the
invention,
Weibull distributions are used to model the reliability and time to failure
for the
equipment components.
8
130817
CA 02470687 2004-06-10
Figs. 4-5 are graphs illustrating the Weibull curves derived for two equipment
components using a base equipment age. Individual data points on the graphs
represent the age data of the components. The solid line represents the
Weibull
estimate for the probability of failure (y axis) at the corresponding age in
days (x
axis). The dotted lines on either side of the solid line represent the 95%
confidence
bounds for the range of probabilities for that age. Both x and y-axes are
represented
on a 1og10 scale. Visual inspection of Figs 6-7 indicate that the data points
follow the
Weibull curve more closely after the first year. For this speciific data set,
there is a
larger deviation from the Weibull estimate for failures within a year.
It must be appreciated that the Weibull curves of Figs 4-5 are obtained using
only the
age of the equipment component as a basis for life estimation. However, usage
pattern data for equipment components, for example, mileage, hours in use,
cycles
and starts are generally not similar across all equipment components. Usage
pattern
data vary considerably over time and across equipment components. Similarly,
as
mentioned above, other factors may contribute to the rate of wear of equipment
components, such as operation under abnormal conditions, variations in
maintenance
practices, or variations of environmental conditions.
Fig. 6 is a table illustrating the goodness of fit of the graphs in Figs.4-5.
The metrics
used are the sum squared errors (SSE) and the percentage of data points
located
within the calculated 95% confidence limits. The SSIE is a cumulative measure
of
how much distance there was between each individual data point and the Weibull
model.
Figs 7-8 are graphs illustrating the Weibull curves derived for two equipment
components based on the unified age adjustment value.. The graphs indicate
that the
Weibull fit produced for the age adjustment value of both the components
provide a
closer, smoother fit than the ones for the base age Weibull fit derived in
Figs. 4-5. A
Weibull fitting that is more accurate has tighter confidence bounds and is a
better
predictor of life remaining for a component. This suggests that the adjusted
equipment age is a better predictor and provides a more refined estimate of
the
9
130817
CA 02470687 2004-06-10
equipment age. This can in turn improve the explanation of historical failures
of
equipment components.
Fig. 9 is a table illustrating the comparison of the goodness of fit of the
graphs of
Figs. 4-5 vs. the graphs in Figs.7-8. The stress variable represents the wear
of the
equipment component. The results from the table also indicate that the Weibull
fit
produced based on the unified age adjustment value of both the equipment
components provide a closer, smoother fit than the ones for the base age
Weibull fit.
The foregoing flow charts and block diagrams of the invention show the
functionality
and operation of the equipment life estimation system 30. In this regard, each
block/component represents a module, segment, or portion of code, which
comprises
one or more executable instructions for implementing the specified logical
function(s). It should also be noted that in some alternative implementations,
the
functions noted in the blocks may occur out of the order noted in the figures
or, for
example, may in fact be executed substantially concurrently or in the reverse
order,
depending upon the functionality involved. Also, one of ordinary skill in the
art will
recognize that additional blocks may be added. Furthermore, the functions can
be
implemented in programming languages such as C++ or JAVA; however, other
languages can be used such as Perl, JavaScript and Visual Basic.
The various embodiments described above comprise an ordered listing of
executable
instructions for implementing logical functions. The ordered listing can be
embodied
in any computer-readable medium for use by or in connection with a computer-
based
system that can retrieve the instructions and execute them. In the context of
the
application, the computer-readable medium can be any means that can contain,
store,
communicate, propagate, transmit or transport the instructions. The computer
readable medium can be an electronic, a magnetic, an optical, an
electromagnetic, or
an infrared system, apparatus, or device. An illustrative, but non-exhaustive
list of
computer-readable mediums can include an electrical connection (electronic)
having
one or more wires, a portable computer diskette (magnetic), a random access
memory
(RAM) (magnetic), a read-only memory (ROM) (magnetic), an erasable
130817
CA 02470687 2004-06-10
programmable read-only memory (EPROM or Flash memory) (magnetic), an optical
fiber (optical), and a portable compact disc read-only memory (CDROM)
(optical).
Note that the computer readable medium may comprise paper or another suitable
medium upon which the instructions are printed. For instance, the instructions
can be
electronically captured via optical scanning of the paper or other medium,
then
compiled, interpreted or otherwise processed in a suitable manner if
necessary, and
then stored in a computer memory.
The embodiments described above have several advantages, including the ability
to
integrate heterogeneous information sources that are potential parameters in
the
estimation of the remaining equipment life. Also, the VVeibull distributions
based on
the unified age adjustment value indicate a closer and smoother Weibull
distribution
fit. A closer fit, in turn, reduces the variability of the Weibull curve and
provides a
better, more refined life estimate for equipment components.
It is apparent that there has been provided, a method, system and computer
product
for estimating a remaining equipment life based on a plurality of parameters.
While
the invention has been particularly shown and described in conjunction with a
preferred embodiment thereof, it will be appreciated thavt variations and
modifications
can be effected by a person of ordinary skill in the art without departing
from the
scope of the invention.
11
